Aticaprant in MDD: target engagement held, clinical efficacy did not

What was tried

Janssen ran a Phase 3 randomized, double-blind, placebo-controlled study (NCT06635135) of adjunctive aticaprant for relapse prevention in major depressive disorder with moderate-to-severe anhedonia. Patients were stabilized on aticaprant added to an antidepressant during open-label treatment, then randomized to continue aticaprant or switch to placebo, with time to first relapse as the primary endpoint. The study started in September 2024, enrolled 47 patients, and was terminated in 2025 for insufficient efficacy. Aticaprant (JNJ-67953964) is a selective kappa-opioid receptor antagonist. It belonged to the broader Phase 3 VENTURA program, which Johnson & Johnson discontinued in March 2025, citing insufficient efficacy in the target population and no new safety signals (Johnson & Johnson, March 2025).

The biological hypothesis

The kappa-opioid receptor and its endogenous ligand dynorphin regulate aversion and reward, and the system is engaged by stress. Blocking the receptor was hypothesized to restore reward processing and relieve anhedonia, a symptom that monoamine antidepressants treat poorly (Browne et al., 2022). The hypothesis had unusually strong human proof of mechanism. In the NIMH fast-fail study, aticaprant 10 mg increased functional MRI ventral striatum activation during reward anticipation, with an effect size of 0.58 (95% CI 0.13 to 0.99, p<0.01) against placebo (Krystal et al., 2020). That is a clean, target-engaged, brain-level readout of the intended pharmacology.

What actually happened

Target engagement held. Clinical efficacy did not. The relapse-prevention study terminated before it accrued enough events to estimate a median time to relapse, and Johnson & Johnson ended the full VENTURA program for insufficient efficacy. The drug was safe and well tolerated throughout. The same pattern appeared in the only other late-stage kappa-opioid antagonist depression program. Neumora's navacaprant missed its primary HAMD-17 endpoint in the full Phase 2 population, with a least-squares mean difference of -1.7 points (95% CI -3.82 to 0.42, p=0.12), and was positive only in a moderate-to-severe subgroup (Mathew et al., 2025). Its Phase 3 KOASTAL-1 study then failed to separate from placebo on the MADRS scale (Neumora, January 2025). Two independent antagonists, two pivotal-stage efficacy failures.

Failure mechanism, best guess

The most likely explanation is a proof-of-mechanism-to-clinic gap, not a wrong target. Kappa-opioid antagonism reliably moves a reward-circuit biomarker, but the translation from striatal activation to a durable change on clinical depression scales is weak. The early positive signals were fragile. Aticaprant's clinical case leaned on the fast-fail biomarker, and navacaprant's leaned on a subgroup analysis after a missed primary. Tested at scale on standard clinical endpoints, the effect regressed toward zero. Anhedonia may be necessary but not sufficient as an entry criterion, and the relapse-prevention design adds its own hurdle. Enriching for stabilized responders during open-label treatment leaves a maintenance population in which drug and placebo are hard to separate.

How to prevent this next time

The recurring error was treating a target-engagement biomarker and a favorable subgroup as evidence of clinical efficacy. Two quantitative tools would have calibrated expectations.

First, an explicit power and effect-size audit anchored to the clinical scale, not the biomarker. A 2-point MADRS difference against a standard deviation near 8 requires about 252 patients per arm at 80% power. A clinically meaningful 4-point difference requires about 63 per arm. A 47-patient relapse-prevention study could only detect a very large effect, so its termination for insufficient efficacy says more about the design than the molecule.

Second, a Bayesian predictive probability of success that propagated the uncertainty in the proof-of-mechanism and Phase 2 data rather than the point estimates.

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

With a prior set to the base rate for central nervous system programs, which sit below the 13.8% overall Phase 1 likelihood of approval reported by Wong et al., 2019, and a likelihood built from a missed Phase 2 primary, that integral would have returned a sobering number well before three pivotal studies were committed. A biomarker-enrichment strategy that linked the reward signal to individual clinical response, validated prospectively, would have been the bridge the field skipped.

The single highest leverage change would have been to require a prospectively validated link between the reward-circuit biomarker and clinical response before launching pivotal trials, rather than treating target engagement as a surrogate for efficacy.

What this means for similar programs

The Open Targets association between OPRK1 and major depressive disorder is 0.57, carried almost entirely by clinical evidence (0.92) rather than genetics (literature component 0.28) (Open Targets Platform). That profile, strong clinical interest and thin genetic support, is exactly where a reproducible biomarker can lead a field into serial pivotal failures. The Claidex Mechanism Risk Score for OPRK1 is now 39 of 100 (yellow band), reflecting one late-phase efficacy failure on a heavily drugged receptor with moderate genetic anchoring. New kappa-opioid programs in mood disorders should expect a high bar, a validated predictive biomarker, a clinical-scale-powered design, and an entry population defined by more than anhedonia alone.

Open questions

Will the full VENTURA dataset, which Johnson & Johnson has said it will present, show any responsive subpopulation? Is the reward-circuit biomarker predictive of clinical response at the individual level, or only at the group level? And does kappa-opioid antagonism have a place in a non-MDD indication, such as substance use or a more narrowly defined anhedonic phenotype, where the company has signaled it may continue?

Sources

1. Johnson & Johnson. Statement on VENTURA Program. March 2025. https://www.jnj.com/media-center/press-releases/johnson-johnson-statement-on-ventura-program.

2. ClinicalTrials.gov. NCT06635135, results record. National Library of Medicine. https://clinicaltrials.gov/study/NCT06635135?tab=results.

3. Krystal AD, Pizzagalli DA, Smoski M, Mathew SJ, Nurnberger J, Lisanby SH, et al. A randomized proof-of-mechanism trial applying the fast-fail approach to evaluating kappa-opioid antagonism as a treatment for anhedonia. Nat Med. 2020, 26(5), 760-768.

4. Mathew SJ, Cutler AJ, Visitacion NC, Gold M, Yuan J, Aurora B. Navacaprant, a Novel and Highly Selective Kappa Opioid Receptor Antagonist, in Adults With Major Depressive Disorder. J Clin Psychopharmacol. 2025, 45(3), 267-276.

5. Browne CA, Wulf H, Lucki I. Kappa Opioid Receptors in the Pathology and Treatment of Major Depressive Disorder. Handb Exp Pharmacol. 2022, 271, 493-524.

6. Wong CH, Siah KW, Lo AW. Estimation of clinical trial success rates and related parameters. Biostatistics. 2019, 20(2), 273-286.